Deep Super-Resolution Network for Single Image Super-Resolution with Realistic Degradations

Umer, Rao Muhammad; Foresti, Gian Luca; Micheloni, Christian

doi:10.1145/3349801.3349823

Cited by 15 publications

(9 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although there are existing SR methods [7], [10], and [17] by handling multiple degradation settings, but they are the non-blind SR techniques with deep feed-forward networks, while our proposed method is a blind SR GANbased approach. So, the comparison with [7], [10], and [17] methods is not fair. We run all the original source codes and trained models by the default parameters settings for the comparison.…”

Section: E Comparison With the State-of-art Methodsmentioning

confidence: 99%

See 1 more Smart Citation

A Deep Residual Star Generative Adversarial Network for multi-domain Image Super-Resolution

Umer¹,

Munir²,

Micheloni³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Recently, most of state-of-the-art single image superresolution (SISR) methods have attained impressive performance by using deep convolutional neural networks (DCNNs). The existing SR methods have limited performance due to a fixed degradation settings, i.e. usually a bicubic downscaling of lowresolution (LR) image. However, in real-world settings, the LR degradation process is unknown which can be bicubic LR, bilinear LR, nearest-neighbor LR, or real LR. Therefore, most SR methods are ineffective and inefficient in handling more than one degradation settings within a single network. To handle the multiple degradation, i.e. refers to multi-domain image super-resolution, we propose a deep Super-Resolution Residual StarGAN (SR 2 * GAN), a novel and scalable approach that superresolves the LR images for the multiple LR domains using only a single model. The proposed scheme is trained in a StarGAN like network topology with a single generator and discriminator networks. We demonstrate the effectiveness of our proposed approach in quantitative and qualitative experiments compared to other state-of-the-art methods.

show abstract

Section: E Comparison With the State-of-art Methodsmentioning

confidence: 99%

“…Lim et al [5] proposed an enhanced deep SR (EDSR) network by taking advantage of the residual learning. In [17], the authors proposed SRWDNet to solve the joint deblurring and superresolution task by following the realistic degradation. In [3], the authors proposed the ISRResCNet to solve the SISR in an iterative manner.…”

Section: Related Workmentioning

confidence: 99%

A Deep Residual Star Generative Adversarial Network for multi-domain Image Super-Resolution

Umer¹,

Munir²,

Micheloni³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…NLOS ): Since the signal power of the NLOS channels are generally weaker than LOS channel, the deconvolution module is used to learn the parameters by exploiting the learnable Wiener Filtering layer [41]. In case of multiple regularization kernels, we formulate the NLOS channels objective function as:…”

Section: D) Convolution Layer (Cmentioning

confidence: 99%

“…Inspired by [41], the optimization problem ( 27) can be solved by Wiener deconvolution technique. To learn a more suitable regularization kernel and parameters in (27), the Wiener filtering layer is developed as follows…”

Section: D) Convolution Layer (Cmentioning

confidence: 99%

Offset Learning Based Channel Estimation for Intelligent Reflecting Surface-Assisted Indoor Communication

Chen

Tang

Zhang

et al. 2022

IEEE J. Sel. Top. Signal Process.

View full text Add to dashboard Cite

The emerging intelligent reflecting surface (IRS) can significantly improve the system capacity, and it has been regarded as a promising technology for the beyond fifth-generation (B5G) communications. For IRS-assisted multiple input multiple output (MIMO) systems, accurate channel estimation is a critical challenge. This severely restricts practical applications, particularly for resource-limited indoor scenario as it contains numerous scatterers and parameters to be estimated, while the number of pilots is limited. Prior art tackles these issues and associated optimization using mathematical-based statistical approaches, but are difficult to solve as the number of scatterers increase. To estimate the indoor channels with an affordable piloting overhead, we propose an offset learning (OL)-based neural network for channel estimation. The proposed OL-based estimator can dynamically trace the channel state information (CSI) without any prior knowledge of the IRS-assisted channel structure as well as indoor statistics. In addition, inspired by the powerful learning capability of convolutional neural network (CNN), CNN-based inversion blocks are developed in the offset estimation module to build the offset estimation operator. Numerical results show that the proposed OL-based estimator can achieve more accurate indoor

show abstract

“…Zhang et al [11] proposed a deep plug-and-play Super-Resolution method for arbitrary blur kernels by following the multiple degradation. In [12], the authors proposed SRWDNet to solve the joint deblurring and super-resolution task by following the realistic degradation. The above methods are deep or wide CNN networks to learn non-linear mapping from LR to HR with a large number of training samples, while neglecting the image acquisition process.…”

Section: Related Workmentioning

confidence: 99%

Deep Iterative Residual Convolutional Network for Single Image Super-Resolution

Umer¹,

Foresti²,

Micheloni³

2020

Preprint

Self Cite

View full text Add to dashboard Cite

Deep convolutional neural networks (CNNs) have recently achieved great success for single image super-resolution (SISR) task due to their powerful feature representation capabilities. The most recent deep learning based SISR methods focus on designing deeper / wider models to learn the non-linear mapping between low-resolution (LR) inputs and high-resolution (HR) outputs. These existing SR methods do not take into account the image observation (physical) model and thus require a large number of network's trainable parameters with a great volume of training data. To address these issues, we propose a deep Iterative Super-Resolution Residual Convolutional Network (ISRResCNet) that exploits the powerful image regularization and large-scale optimization techniques by training the deep network in an iterative manner with a residual learning approach. Extensive experimental results on various super-resolution benchmarks demonstrate that our method with a few trainable parameters improves the results for different scaling factors in comparison with the state-of-art methods.

show abstract

Deep Super-Resolution Network for Single Image Super-Resolution with Realistic Degradations

Cited by 15 publications

References 35 publications

A Deep Residual Star Generative Adversarial Network for multi-domain Image Super-Resolution

A Deep Residual Star Generative Adversarial Network for multi-domain Image Super-Resolution

Offset Learning Based Channel Estimation for Intelligent Reflecting Surface-Assisted Indoor Communication

Deep Iterative Residual Convolutional Network for Single Image Super-Resolution

Contact Info

Product

Resources

About